This patch introduces support for the BPF_F_CPU flag in percpu_array maps
to allow updating or looking up values for specified CPUs or for all CPUs
with a single value.

This enhancement enables:

* Efficient update of all CPUs using a single value when cpu == (u32)~0.
* Targeted update or lookup for a specified CPU otherwise.

The flag is passed via:

* map_flags in bpf_percpu_array_update() along with embedded cpu field.
* elem_flags in generic_map_update_batch() along with separated cpu field.

Signed-off-by: Leon Hwang <leon.hwang@linux.dev>
---
 include/linux/bpf.h            |  3 +-
 include/uapi/linux/bpf.h       |  7 +++++
 kernel/bpf/arraymap.c          | 54 ++++++++++++++++++++++++++--------
 kernel/bpf/syscall.c           | 52 ++++++++++++++++++++------------
 tools/include/uapi/linux/bpf.h |  7 +++++
 5 files changed, 90 insertions(+), 33 deletions(-)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index f9cd2164ed23..faee5710e913 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -2671,7 +2671,8 @@ int map_set_for_each_callback_args(struct bpf_verifier_env *env,
 				   struct bpf_func_state *callee);
 
 int bpf_percpu_hash_copy(struct bpf_map *map, void *key, void *value);
-int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value);
+int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value,
+			  u64 flags);
 int bpf_percpu_hash_update(struct bpf_map *map, void *key, void *value,
 			   u64 flags);
 int bpf_percpu_array_update(struct bpf_map *map, void *key, void *value,
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index 233de8677382..4cad3de6899d 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -1372,6 +1372,12 @@ enum {
 	BPF_NOEXIST	= 1, /* create new element if it didn't exist */
 	BPF_EXIST	= 2, /* update existing element */
 	BPF_F_LOCK	= 4, /* spin_lock-ed map_lookup/map_update */
+	BPF_F_CPU	= 8, /* map_update for percpu_array */
+};
+
+enum {
+	/* indicate updating value across all CPUs for percpu maps. */
+	BPF_ALL_CPUS	= (__u32)~0,
 };
 
 /* flags for BPF_MAP_CREATE command */
@@ -1549,6 +1555,7 @@ union bpf_attr {
 		__u32		map_fd;
 		__u64		elem_flags;
 		__u64		flags;
+		__u32		cpu;
 	} batch;
 
 	struct { /* anonymous struct used by BPF_PROG_LOAD command */
diff --git a/kernel/bpf/arraymap.c b/kernel/bpf/arraymap.c
index 3d080916faf9..d333663cbe71 100644
--- a/kernel/bpf/arraymap.c
+++ b/kernel/bpf/arraymap.c
@@ -295,17 +295,24 @@ static void *percpu_array_map_lookup_percpu_elem(struct bpf_map *map, void *key,
 	return per_cpu_ptr(array->pptrs[index & array->index_mask], cpu);
 }
 
-int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value)
+int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value, u64 flags)
 {
 	struct bpf_array *array = container_of(map, struct bpf_array, map);
 	u32 index = *(u32 *)key;
 	void __percpu *pptr;
-	int cpu, off = 0;
-	u32 size;
+	u32 size, cpu;
+	int off = 0;
 
 	if (unlikely(index >= array->map.max_entries))
 		return -ENOENT;
 
+	cpu = (u32)(flags >> 32);
+	flags &= (u32)~0;
+	if (unlikely(flags > BPF_F_CPU))
+		return -EINVAL;
+	if (unlikely((flags & BPF_F_CPU) && cpu >= num_possible_cpus()))
+		return -E2BIG;
+
 	/* per_cpu areas are zero-filled and bpf programs can only
 	 * access 'value_size' of them, so copying rounded areas
 	 * will not leak any kernel data
@@ -313,10 +320,15 @@ int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value)
 	size = array->elem_size;
 	rcu_read_lock();
 	pptr = array->pptrs[index & array->index_mask];
-	for_each_possible_cpu(cpu) {
-		copy_map_value_long(map, value + off, per_cpu_ptr(pptr, cpu));
-		check_and_init_map_value(map, value + off);
-		off += size;
+	if (flags & BPF_F_CPU) {
+		copy_map_value_long(map, value, per_cpu_ptr(pptr, cpu));
+		check_and_init_map_value(map, value);
+	} else {
+		for_each_possible_cpu(cpu) {
+			copy_map_value_long(map, value + off, per_cpu_ptr(pptr, cpu));
+			check_and_init_map_value(map, value + off);
+			off += size;
+		}
 	}
 	rcu_read_unlock();
 	return 0;
@@ -387,13 +399,20 @@ int bpf_percpu_array_update(struct bpf_map *map, void *key, void *value,
 	struct bpf_array *array = container_of(map, struct bpf_array, map);
 	u32 index = *(u32 *)key;
 	void __percpu *pptr;
-	int cpu, off = 0;
-	u32 size;
+	u32 size, cpu;
+	int off = 0;
 
-	if (unlikely(map_flags > BPF_EXIST))
+	cpu = (u32)(map_flags >> 32);
+	map_flags &= (u32)~0;
+	if (unlikely(map_flags > BPF_F_CPU))
 		/* unknown flags */
 		return -EINVAL;
 
+	if (unlikely((map_flags & BPF_F_CPU) && cpu != BPF_ALL_CPUS &&
+		     cpu >= num_possible_cpus()))
+		/* invalid cpu */
+		return -E2BIG;
+
 	if (unlikely(index >= array->map.max_entries))
 		/* all elements were pre-allocated, cannot insert a new one */
 		return -E2BIG;
@@ -411,10 +430,19 @@ int bpf_percpu_array_update(struct bpf_map *map, void *key, void *value,
 	size = array->elem_size;
 	rcu_read_lock();
 	pptr = array->pptrs[index & array->index_mask];
-	for_each_possible_cpu(cpu) {
-		copy_map_value_long(map, per_cpu_ptr(pptr, cpu), value + off);
+	if ((map_flags & BPF_F_CPU) && cpu != BPF_ALL_CPUS) {
+		copy_map_value_long(map, per_cpu_ptr(pptr, cpu), value);
 		bpf_obj_free_fields(array->map.record, per_cpu_ptr(pptr, cpu));
-		off += size;
+	} else {
+		for_each_possible_cpu(cpu) {
+			copy_map_value_long(map, per_cpu_ptr(pptr, cpu), value + off);
+			/* same user-provided value is used if BPF_F_CPU is specified,
+			 * otherwise value is an array of per-cpu values.
+			 */
+			if (!(map_flags & BPF_F_CPU))
+				off += size;
+			bpf_obj_free_fields(array->map.record, per_cpu_ptr(pptr, cpu));
+		}
 	}
 	rcu_read_unlock();
 	return 0;
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index e63039817af3..0b49610d5939 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -129,8 +129,11 @@ bool bpf_map_write_active(const struct bpf_map *map)
 	return atomic64_read(&map->writecnt) != 0;
 }
 
-static u32 bpf_map_value_size(const struct bpf_map *map)
+static u32 bpf_map_value_size(const struct bpf_map *map, u64 flags)
 {
+	if ((flags & BPF_F_CPU) && map->map_type == BPF_MAP_TYPE_PERCPU_ARRAY)
+		return round_up(map->value_size, 8);
+
 	if (map->map_type == BPF_MAP_TYPE_PERCPU_HASH ||
 	    map->map_type == BPF_MAP_TYPE_LRU_PERCPU_HASH ||
 	    map->map_type == BPF_MAP_TYPE_PERCPU_ARRAY ||
@@ -312,7 +315,7 @@ static int bpf_map_copy_value(struct bpf_map *map, void *key, void *value,
 	    map->map_type == BPF_MAP_TYPE_LRU_PERCPU_HASH) {
 		err = bpf_percpu_hash_copy(map, key, value);
 	} else if (map->map_type == BPF_MAP_TYPE_PERCPU_ARRAY) {
-		err = bpf_percpu_array_copy(map, key, value);
+		err = bpf_percpu_array_copy(map, key, value, flags);
 	} else if (map->map_type == BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE) {
 		err = bpf_percpu_cgroup_storage_copy(map, key, value);
 	} else if (map->map_type == BPF_MAP_TYPE_STACK_TRACE) {
@@ -1662,7 +1665,8 @@ static int map_lookup_elem(union bpf_attr *attr)
 	if (CHECK_ATTR(BPF_MAP_LOOKUP_ELEM))
 		return -EINVAL;
 
-	if (attr->flags & ~BPF_F_LOCK)
+	if (((u32)attr->flags & ~(BPF_F_LOCK | BPF_F_CPU)) ||
+		(!((u32)attr->flags & BPF_F_CPU) && attr->flags >> 32))
 		return -EINVAL;
 
 	CLASS(fd, f)(attr->map_fd);
@@ -1680,7 +1684,7 @@ static int map_lookup_elem(union bpf_attr *attr)
 	if (IS_ERR(key))
 		return PTR_ERR(key);
 
-	value_size = bpf_map_value_size(map);
+	value_size = bpf_map_value_size(map, attr->flags);
 
 	err = -ENOMEM;
 	value = kvmalloc(value_size, GFP_USER | __GFP_NOWARN);
@@ -1749,7 +1753,7 @@ static int map_update_elem(union bpf_attr *attr, bpfptr_t uattr)
 		goto err_put;
 	}
 
-	value_size = bpf_map_value_size(map);
+	value_size = bpf_map_value_size(map, attr->flags);
 	value = kvmemdup_bpfptr(uvalue, value_size);
 	if (IS_ERR(value)) {
 		err = PTR_ERR(value);
@@ -1941,19 +1945,25 @@ int generic_map_update_batch(struct bpf_map *map, struct file *map_file,
 {
 	void __user *values = u64_to_user_ptr(attr->batch.values);
 	void __user *keys = u64_to_user_ptr(attr->batch.keys);
-	u32 value_size, cp, max_count;
+	u32 value_size, cp, max_count, cpu = attr->batch.cpu;
+	u64 elem_flags = attr->batch.elem_flags;
 	void *key, *value;
 	int err = 0;
 
-	if (attr->batch.elem_flags & ~BPF_F_LOCK)
+	if (elem_flags & ~(BPF_F_LOCK | BPF_F_CPU))
 		return -EINVAL;
 
-	if ((attr->batch.elem_flags & BPF_F_LOCK) &&
+	if ((elem_flags & BPF_F_LOCK) &&
 	    !btf_record_has_field(map->record, BPF_SPIN_LOCK)) {
 		return -EINVAL;
 	}
 
-	value_size = bpf_map_value_size(map);
+	if ((elem_flags & BPF_F_CPU) &&
+		map->map_type != BPF_MAP_TYPE_PERCPU_ARRAY)
+		return -EINVAL;
+
+	value_size = bpf_map_value_size(map, elem_flags);
+	elem_flags |= ((u64)cpu) << 32;
 
 	max_count = attr->batch.count;
 	if (!max_count)
@@ -1979,8 +1989,7 @@ int generic_map_update_batch(struct bpf_map *map, struct file *map_file,
 		    copy_from_user(value, values + cp * value_size, value_size))
 			break;
 
-		err = bpf_map_update_value(map, map_file, key, value,
-					   attr->batch.elem_flags);
+		err = bpf_map_update_value(map, map_file, key, value, elem_flags);
 
 		if (err)
 			break;
@@ -2004,18 +2013,24 @@ int generic_map_lookup_batch(struct bpf_map *map,
 	void __user *ubatch = u64_to_user_ptr(attr->batch.in_batch);
 	void __user *values = u64_to_user_ptr(attr->batch.values);
 	void __user *keys = u64_to_user_ptr(attr->batch.keys);
+	u32 value_size, cp, max_count, cpu = attr->batch.cpu;
 	void *buf, *buf_prevkey, *prev_key, *key, *value;
-	u32 value_size, cp, max_count;
+	u64 elem_flags = attr->batch.elem_flags;
 	int err;
 
-	if (attr->batch.elem_flags & ~BPF_F_LOCK)
+	if (elem_flags & ~(BPF_F_LOCK | BPF_F_CPU))
 		return -EINVAL;
 
-	if ((attr->batch.elem_flags & BPF_F_LOCK) &&
+	if ((elem_flags & BPF_F_LOCK) &&
 	    !btf_record_has_field(map->record, BPF_SPIN_LOCK))
 		return -EINVAL;
 
-	value_size = bpf_map_value_size(map);
+	if ((elem_flags & BPF_F_CPU) &&
+		map->map_type != BPF_MAP_TYPE_PERCPU_ARRAY)
+		return -EINVAL;
+
+	value_size = bpf_map_value_size(map, elem_flags);
+	elem_flags |= ((u64)cpu) << 32;
 
 	max_count = attr->batch.count;
 	if (!max_count)
@@ -2049,8 +2064,7 @@ int generic_map_lookup_batch(struct bpf_map *map,
 		rcu_read_unlock();
 		if (err)
 			break;
-		err = bpf_map_copy_value(map, key, value,
-					 attr->batch.elem_flags);
+		err = bpf_map_copy_value(map, key, value, elem_flags);
 
 		if (err == -ENOENT)
 			goto next_key;
@@ -2137,7 +2151,7 @@ static int map_lookup_and_delete_elem(union bpf_attr *attr)
 		goto err_put;
 	}
 
-	value_size = bpf_map_value_size(map);
+	value_size = bpf_map_value_size(map, attr->flags);
 
 	err = -ENOMEM;
 	value = kvmalloc(value_size, GFP_USER | __GFP_NOWARN);
@@ -5457,7 +5471,7 @@ static int bpf_task_fd_query(const union bpf_attr *attr,
 	return err;
 }
 
-#define BPF_MAP_BATCH_LAST_FIELD batch.flags
+#define BPF_MAP_BATCH_LAST_FIELD batch.cpu
 
 #define BPF_DO_BATCH(fn, ...)			\
 	do {					\
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index 233de8677382..4cad3de6899d 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -1372,6 +1372,12 @@ enum {
 	BPF_NOEXIST	= 1, /* create new element if it didn't exist */
 	BPF_EXIST	= 2, /* update existing element */
 	BPF_F_LOCK	= 4, /* spin_lock-ed map_lookup/map_update */
+	BPF_F_CPU	= 8, /* map_update for percpu_array */
+};
+
+enum {
+	/* indicate updating value across all CPUs for percpu maps. */
+	BPF_ALL_CPUS	= (__u32)~0,
 };
 
 /* flags for BPF_MAP_CREATE command */
@@ -1549,6 +1555,7 @@ union bpf_attr {
 		__u32		map_fd;
 		__u64		elem_flags;
 		__u64		flags;
+		__u32		cpu;
 	} batch;
 
 	struct { /* anonymous struct used by BPF_PROG_LOAD command */
-- 
2.50.1

This patch adds libbpf support for the BPF_F_CPU flag in percpu_array maps,
introducing the following APIs:

1. bpf_map_update_elem_opts(): update with struct bpf_map_update_elem_opts
2. bpf_map_lookup_elem_opts(): lookup with struct bpf_map_lookup_elem_opts
3. bpf_map__update_elem_opts(): high-level wrapper with input validation
4. bpf_map__lookup_elem_opts(): high-level wrapper with input validation

Behavior:

* If opts->cpu == (u32)~0, the update is applied to all CPUs.
* Otherwise, it applies only to the specified CPU.
* Lookup APIs retrieve values from the target CPU when BPF_F_CPU is used.

Signed-off-by: Leon Hwang <leon.hwang@linux.dev>
---
 tools/lib/bpf/bpf.c           | 23 ++++++++++++++
 tools/lib/bpf/bpf.h           | 36 +++++++++++++++++++++-
 tools/lib/bpf/libbpf.c        | 56 +++++++++++++++++++++++++++++++----
 tools/lib/bpf/libbpf.h        | 53 ++++++++++++++++++++++++++++-----
 tools/lib/bpf/libbpf.map      |  5 ++++
 tools/lib/bpf/libbpf_common.h | 14 +++++++++
 6 files changed, 173 insertions(+), 14 deletions(-)

diff --git a/tools/lib/bpf/bpf.c b/tools/lib/bpf/bpf.c
index ab40dbf9f020..8061093d84f9 100644
--- a/tools/lib/bpf/bpf.c
+++ b/tools/lib/bpf/bpf.c
@@ -402,6 +402,17 @@ int bpf_map_update_elem(int fd, const void *key, const void *value,
 	return libbpf_err_errno(ret);
 }
 
+int bpf_map_update_elem_opts(int fd, const void *key, const void *value,
+			     const struct bpf_map_update_elem_opts *opts)
+{
+	__u64 flags;
+	__u32 cpu;
+
+	cpu = OPTS_GET(opts, cpu, 0);
+	flags = ((__u64) cpu) << 32 | OPTS_GET(opts, flags, 0);
+	return bpf_map_update_elem(fd, key, value, flags);
+}
+
 int bpf_map_lookup_elem(int fd, const void *key, void *value)
 {
 	const size_t attr_sz = offsetofend(union bpf_attr, flags);
@@ -433,6 +444,17 @@ int bpf_map_lookup_elem_flags(int fd, const void *key, void *value, __u64 flags)
 	return libbpf_err_errno(ret);
 }
 
+int bpf_map_lookup_elem_opts(int fd, const void *key, void *value,
+			     const struct bpf_map_lookup_elem_opts *opts)
+{
+	__u64 flags;
+	__u32 cpu;
+
+	cpu = OPTS_GET(opts, cpu, 0);
+	flags = ((__u64) cpu) << 32 | OPTS_GET(opts, flags, 0);
+	return bpf_map_lookup_elem_flags(fd, key, value, flags);
+}
+
 int bpf_map_lookup_and_delete_elem(int fd, const void *key, void *value)
 {
 	const size_t attr_sz = offsetofend(union bpf_attr, flags);
@@ -542,6 +564,7 @@ static int bpf_map_batch_common(int cmd, int fd, void  *in_batch,
 	attr.batch.count = *count;
 	attr.batch.elem_flags  = OPTS_GET(opts, elem_flags, 0);
 	attr.batch.flags = OPTS_GET(opts, flags, 0);
+	attr.batch.cpu = OPTS_GET(opts, cpu, 0);
 
 	ret = sys_bpf(cmd, &attr, attr_sz);
 	*count = attr.batch.count;
diff --git a/tools/lib/bpf/bpf.h b/tools/lib/bpf/bpf.h
index 7252150e7ad3..d0ab18b50294 100644
--- a/tools/lib/bpf/bpf.h
+++ b/tools/lib/bpf/bpf.h
@@ -163,12 +163,42 @@ LIBBPF_API int bpf_map_delete_elem_flags(int fd, const void *key, __u64 flags);
 LIBBPF_API int bpf_map_get_next_key(int fd, const void *key, void *next_key);
 LIBBPF_API int bpf_map_freeze(int fd);
 
+/**
+ * @brief **bpf_map_update_elem_opts** allows for updating map with options.
+ *
+ * @param fd BPF map file descriptor
+ * @param key pointer to key
+ * @param value pointer to value
+ * @param opts options for configuring the way to update map
+ * @return 0, on success; negative error code, otherwise (errno is also set to
+ * the error code)
+ */
+LIBBPF_API int bpf_map_update_elem_opts(int fd, const void *key, const void *value,
+					const struct bpf_map_update_elem_opts *opts);
+
+/**
+ * @brief **bpf_map_lookup_elem_opts** allows for looking up the value with
+ * options.
+ *
+ * @param fd BPF map file descriptor
+ * @param key pointer to key
+ * @param value pointer to value
+ * @param opts options for configuring the way to lookup map
+ * @return 0, on success; negative error code, otherwise (errno is also set to
+ * the error code)
+ */
+LIBBPF_API int bpf_map_lookup_elem_opts(int fd, const void *key, void *value,
+					const struct bpf_map_lookup_elem_opts *opts);
+
+
 struct bpf_map_batch_opts {
 	size_t sz; /* size of this struct for forward/backward compatibility */
 	__u64 elem_flags;
 	__u64 flags;
+	__u32 cpu;
+	size_t:0;
 };
-#define bpf_map_batch_opts__last_field flags
+#define bpf_map_batch_opts__last_field cpu
 
 
 /**
@@ -286,6 +316,10 @@ LIBBPF_API int bpf_map_lookup_and_delete_batch(int fd, void *in_batch,
  *    Update spin_lock-ed map elements. This must be
  *    specified if the map value contains a spinlock.
  *
+ * **BPF_F_CPU**
+ *    As for percpu map, update value across all CPUs if **opts->cpu** is
+ *    (__u32)~0, or on specified CPU otherwise.
+ *
  * @param fd BPF map file descriptor
  * @param keys pointer to an array of *count* keys
  * @param values pointer to an array of *count* values
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index aee36402f0a3..35faedef6ab4 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -10582,7 +10582,8 @@ bpf_object__find_map_fd_by_name(const struct bpf_object *obj, const char *name)
 }
 
 static int validate_map_op(const struct bpf_map *map, size_t key_sz,
-			   size_t value_sz, bool check_value_sz)
+			   size_t value_sz, bool check_value_sz, __u64 flags,
+			   __u32 cpu)
 {
 	if (!map_is_created(map)) /* map is not yet created */
 		return -ENOENT;
@@ -10601,6 +10602,19 @@ static int validate_map_op(const struct bpf_map *map, size_t key_sz,
 	if (!check_value_sz)
 		return 0;
 
+	if (flags & BPF_F_CPU) {
+		if (map->def.type != BPF_MAP_TYPE_PERCPU_ARRAY)
+			return -EINVAL;
+		if (cpu != BPF_ALL_CPUS && cpu >= libbpf_num_possible_cpus())
+			return -E2BIG;
+		if (map->def.value_size != value_sz) {
+			pr_warn("map '%s': unexpected value size %zu provided, expected %u\n",
+				map->name, value_sz, map->def.value_size);
+			return -EINVAL;
+		}
+		return 0;
+	}
+
 	switch (map->def.type) {
 	case BPF_MAP_TYPE_PERCPU_ARRAY:
 	case BPF_MAP_TYPE_PERCPU_HASH:
@@ -10633,32 +10647,62 @@ int bpf_map__lookup_elem(const struct bpf_map *map,
 {
 	int err;
 
-	err = validate_map_op(map, key_sz, value_sz, true);
+	err = validate_map_op(map, key_sz, value_sz, true, 0, 0);
 	if (err)
 		return libbpf_err(err);
 
 	return bpf_map_lookup_elem_flags(map->fd, key, value, flags);
 }
 
+int bpf_map__lookup_elem_opts(const struct bpf_map *map, const void *key,
+			      size_t key_sz, void *value, size_t value_sz,
+			      const struct bpf_map_lookup_elem_opts *opts)
+{
+	__u64 flags = OPTS_GET(opts, flags, 0);
+	__u32 cpu = OPTS_GET(opts, cpu, 0);
+	int err;
+
+	err = validate_map_op(map, key_sz, value_sz, true, flags, cpu);
+	if (err)
+		return libbpf_err(err);
+
+	return bpf_map_lookup_elem_opts(map->fd, key, value, opts);
+}
+
 int bpf_map__update_elem(const struct bpf_map *map,
 			 const void *key, size_t key_sz,
 			 const void *value, size_t value_sz, __u64 flags)
 {
 	int err;
 
-	err = validate_map_op(map, key_sz, value_sz, true);
+	err = validate_map_op(map, key_sz, value_sz, true, 0, 0);
 	if (err)
 		return libbpf_err(err);
 
 	return bpf_map_update_elem(map->fd, key, value, flags);
 }
 
+int bpf_map__update_elem_opts(const struct bpf_map *map, const void *key,
+			      size_t key_sz, const void *value, size_t value_sz,
+			      const struct bpf_map_update_elem_opts *opts)
+{
+	__u64 flags = OPTS_GET(opts, flags, 0);
+	__u32 cpu = OPTS_GET(opts, cpu, 0);
+	int err;
+
+	err = validate_map_op(map, key_sz, value_sz, true, flags, cpu);
+	if (err)
+		return libbpf_err(err);
+
+	return bpf_map_update_elem_opts(map->fd, key, value, opts);
+}
+
 int bpf_map__delete_elem(const struct bpf_map *map,
 			 const void *key, size_t key_sz, __u64 flags)
 {
 	int err;
 
-	err = validate_map_op(map, key_sz, 0, false /* check_value_sz */);
+	err = validate_map_op(map, key_sz, 0, false /* check_value_sz */, 0, 0);
 	if (err)
 		return libbpf_err(err);
 
@@ -10671,7 +10715,7 @@ int bpf_map__lookup_and_delete_elem(const struct bpf_map *map,
 {
 	int err;
 
-	err = validate_map_op(map, key_sz, value_sz, true);
+	err = validate_map_op(map, key_sz, value_sz, true, 0, 0);
 	if (err)
 		return libbpf_err(err);
 
@@ -10683,7 +10727,7 @@ int bpf_map__get_next_key(const struct bpf_map *map,
 {
 	int err;
 
-	err = validate_map_op(map, key_sz, 0, false /* check_value_sz */);
+	err = validate_map_op(map, key_sz, 0, false /* check_value_sz */, 0, 0);
 	if (err)
 		return libbpf_err(err);
 
diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h
index d1cf813a057b..fd4940759bc9 100644
--- a/tools/lib/bpf/libbpf.h
+++ b/tools/lib/bpf/libbpf.h
@@ -1168,13 +1168,7 @@ LIBBPF_API struct bpf_map *bpf_map__inner_map(struct bpf_map *map);
  * @param key pointer to memory containing bytes of the key used for lookup
  * @param key_sz size in bytes of key data, needs to match BPF map definition's **key_size**
  * @param value pointer to memory in which looked up value will be stored
- * @param value_sz size in byte of value data memory; it has to match BPF map
- * definition's **value_size**. For per-CPU BPF maps value size has to be
- * a product of BPF map value size and number of possible CPUs in the system
- * (could be fetched with **libbpf_num_possible_cpus()**). Note also that for
- * per-CPU values value size has to be aligned up to closest 8 bytes for
- * alignment reasons, so expected size is: `round_up(value_size, 8)
- * * libbpf_num_possible_cpus()`.
+ * @param value_sz refer to **bpf_map__lookup_elem_opts()**'s description.
  * @flags extra flags passed to kernel for this operation
  * @return 0, on success; negative error, otherwise
  *
@@ -1185,6 +1179,32 @@ LIBBPF_API int bpf_map__lookup_elem(const struct bpf_map *map,
 				    const void *key, size_t key_sz,
 				    void *value, size_t value_sz, __u64 flags);
 
+/**
+ * @brief **bpf_map__lookup_elem_opts()** allows to lookup BPF map value
+ * corresponding to provided key with options.
+ * @param map BPF map to lookup element in
+ * @param key pointer to memory containing bytes of the key used for lookup
+ * @param key_sz size in bytes of key data, needs to match BPF map definition's **key_size**
+ * @param value pointer to memory in which looked up value will be stored
+ * @param value_sz size in byte of value data memory; it has to match BPF map
+ * definition's **value_size**. For per-CPU BPF maps value size can be
+ * definition's **value_size** if **BPF_F_CPU** is specified in **opts->flags**,
+ * otherwise a product of BPF map value size and number of possible CPUs in the
+ * system (could be fetched with **libbpf_num_possible_cpus()**). Note else that
+ * for per-CPU values value size has to be aligned up to closest 8 bytes for
+ * alignment reasons, so expected size is: `round_up(value_size, 8)
+ * * libbpf_num_possible_cpus()`.
+ * @opts extra options passed to kernel for this operation
+ * @return 0, on success; negative error, otherwise
+ *
+ * **bpf_map__lookup_elem_opts()** is high-level equivalent of
+ * **bpf_map_lookup_elem_opts()** API with added check for key and value size.
+ */
+LIBBPF_API int bpf_map__lookup_elem_opts(const struct bpf_map *map,
+					 const void *key, size_t key_sz,
+					 void *value, size_t value_sz,
+					 const struct bpf_map_lookup_elem_opts *opts);
+
 /**
  * @brief **bpf_map__update_elem()** allows to insert or update value in BPF
  * map that corresponds to provided key.
@@ -1209,6 +1229,25 @@ LIBBPF_API int bpf_map__update_elem(const struct bpf_map *map,
 				    const void *key, size_t key_sz,
 				    const void *value, size_t value_sz, __u64 flags);
 
+/**
+ * @brief **bpf_map__update_elem_opts()** allows to insert or update value in BPF
+ * map that corresponds to provided key with options.
+ * @param map BPF map to insert to or update element in
+ * @param key pointer to memory containing bytes of the key
+ * @param key_sz size in bytes of key data, needs to match BPF map definition's **key_size**
+ * @param value pointer to memory containing bytes of the value
+ * @param value_sz refer to **bpf_map__lookup_elem_opts()**'s description.
+ * @opts extra options passed to kernel for this operation
+ * @return 0, on success; negative error, otherwise
+ *
+ * **bpf_map__update_elem_opts()** is high-level equivalent of
+ * **bpf_map_update_elem_opts()** API with added check for key and value size.
+ */
+LIBBPF_API int bpf_map__update_elem_opts(const struct bpf_map *map,
+					 const void *key, size_t key_sz,
+					 const void *value, size_t value_sz,
+					 const struct bpf_map_update_elem_opts *opts);
+
 /**
  * @brief **bpf_map__delete_elem()** allows to delete element in BPF map that
  * corresponds to provided key.
diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map
index d7bd463e7017..fa415d9f26ce 100644
--- a/tools/lib/bpf/libbpf.map
+++ b/tools/lib/bpf/libbpf.map
@@ -448,4 +448,9 @@ LIBBPF_1.6.0 {
 } LIBBPF_1.5.0;
 
 LIBBPF_1.7.0 {
+	global:
+		bpf_map__lookup_elem_opts;
+		bpf_map__update_elem_opts;
+		bpf_map_lookup_elem_opts;
+		bpf_map_update_elem_opts;
 } LIBBPF_1.6.0;
diff --git a/tools/lib/bpf/libbpf_common.h b/tools/lib/bpf/libbpf_common.h
index 8fe248e14eb6..84ca89ace1be 100644
--- a/tools/lib/bpf/libbpf_common.h
+++ b/tools/lib/bpf/libbpf_common.h
@@ -89,4 +89,18 @@
 		memcpy(&NAME, &___##NAME, sizeof(NAME));		    \
 	} while (0)
 
+struct bpf_map_update_elem_opts {
+	size_t sz; /* size of this struct for forward/backward compatibility */
+	__u64 flags;
+	__u32 cpu;
+	size_t:0;
+};
+
+struct bpf_map_lookup_elem_opts {
+	size_t sz; /* size of this struct for forward/backward compatibility */
+	__u64 flags;
+	__u32 cpu;
+	size_t:0;
+};
+
 #endif /* __LIBBPF_LIBBPF_COMMON_H */
-- 
2.50.1

This patch adds test coverage for the new BPF_F_CPU flag support in
percpu_array maps. The following APIs are exercised:

* bpf_map_update_batch()
* bpf_map_lookup_batch()
* bpf_map_update_elem_opts()
* bpf_map__update_elem_opts()
* bpf_map_lookup_elem_opts()
* bpf_map__lookup_elem_opts()

cd tools/testing/selftests/bpf/
./test_progs -t percpu_alloc/cpu_flag_tests
253/13  percpu_alloc/cpu_flag_tests:OK
253     percpu_alloc:OK
Summary: 1/13 PASSED, 0 SKIPPED, 0 FAILED

Signed-off-by: Leon Hwang <leon.hwang@linux.dev>
---
 .../selftests/bpf/prog_tests/percpu_alloc.c   | 172 ++++++++++++++++++
 .../selftests/bpf/progs/percpu_array_flag.c   |  24 +++
 2 files changed, 196 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/progs/percpu_array_flag.c

diff --git a/tools/testing/selftests/bpf/prog_tests/percpu_alloc.c b/tools/testing/selftests/bpf/prog_tests/percpu_alloc.c
index 343da65864d6..2bf4e6f1d357 100644
--- a/tools/testing/selftests/bpf/prog_tests/percpu_alloc.c
+++ b/tools/testing/selftests/bpf/prog_tests/percpu_alloc.c
@@ -3,6 +3,7 @@
 #include "percpu_alloc_array.skel.h"
 #include "percpu_alloc_cgrp_local_storage.skel.h"
 #include "percpu_alloc_fail.skel.h"
+#include "percpu_array_flag.skel.h"
 
 static void test_array(void)
 {
@@ -115,6 +116,175 @@ static void test_failure(void) {
 	RUN_TESTS(percpu_alloc_fail);
 }
 
+static void test_cpu_flag(void)
+{
+	int map_fd, *keys = NULL, value_size, cpu, i, j, nr_cpus, err;
+	size_t key_sz = sizeof(int), value_sz = sizeof(u64);
+	struct percpu_array_flag *skel;
+	u64 batch = 0, *values = NULL;
+	const u64 value = 0xDEADC0DE;
+	u32 count, max_entries;
+	struct bpf_map *map;
+	LIBBPF_OPTS(bpf_map_lookup_elem_opts, lookup_opts,
+		    .flags = BPF_F_CPU,
+		    .cpu = 0,
+	);
+	LIBBPF_OPTS(bpf_map_update_elem_opts, update_opts,
+		    .flags = BPF_F_CPU,
+		    .cpu = 0,
+	);
+	LIBBPF_OPTS(bpf_map_batch_opts, batch_opts,
+		    .elem_flags = BPF_F_CPU,
+		    .flags = 0,
+	);
+
+	nr_cpus = libbpf_num_possible_cpus();
+	if (!ASSERT_GT(nr_cpus, 0, "libbpf_num_possible_cpus"))
+		return;
+
+	skel = percpu_array_flag__open_and_load();
+	if (!ASSERT_OK_PTR(skel, "percpu_array_flag__open_and_load"))
+		return;
+
+	map = skel->maps.percpu;
+	map_fd = bpf_map__fd(map);
+	max_entries = bpf_map__max_entries(map);
+
+	value_size = value_sz * nr_cpus;
+	values = calloc(max_entries, value_size);
+	if (!ASSERT_OK_PTR(values, "calloc values"))
+		goto out;
+	keys = calloc(max_entries, key_sz);
+	if (!ASSERT_OK_PTR(keys, "calloc keys"))
+		goto out;
+
+	for (i = 0; i < max_entries; i++)
+		keys[i] = i;
+	memset(values, 0, max_entries * value_size);
+
+	batch_opts.cpu = nr_cpus;
+	err = bpf_map_update_batch(map_fd, keys, values, &max_entries, &batch_opts);
+	if (!ASSERT_EQ(err, -E2BIG, "bpf_map_update_batch -E2BIG"))
+		goto out;
+
+	for (cpu = 0; cpu < nr_cpus; cpu++) {
+		memset(values, 0, max_entries * value_size);
+
+		/* clear values on all CPUs */
+		batch_opts.cpu = BPF_ALL_CPUS;
+		batch_opts.elem_flags = BPF_F_CPU;
+		err = bpf_map_update_batch(map_fd, keys, values, &max_entries, &batch_opts);
+		if (!ASSERT_OK(err, "bpf_map_update_batch all cpus"))
+			goto out;
+
+		/* update values on current CPU */
+		for (i = 0; i < max_entries; i++)
+			values[i] = value;
+
+		batch_opts.cpu = cpu;
+		err = bpf_map_update_batch(map_fd, keys, values, &max_entries, &batch_opts);
+		if (!ASSERT_OK(err, "bpf_map_update_batch current cpu"))
+			goto out;
+
+		/* lookup values on current CPU */
+		batch_opts.cpu = cpu;
+		batch_opts.elem_flags = BPF_F_CPU;
+		memset(values, 0, max_entries * value_sz);
+		err = bpf_map_lookup_batch(map_fd, NULL, &batch, keys, values, &count, &batch_opts);
+		if (!ASSERT_TRUE(!err || err == -ENOENT, "bpf_map_lookup_batch current cpu"))
+			goto out;
+
+		for (i = 0; i < max_entries; i++)
+			if (!ASSERT_EQ(values[i], value, "value on current cpu"))
+				goto out;
+
+		/* lookup values on all CPUs */
+		batch_opts.cpu = 0;
+		batch_opts.elem_flags = 0;
+		memset(values, 0, max_entries * value_size);
+		err = bpf_map_lookup_batch(map_fd, NULL, &batch, keys, values, &count, &batch_opts);
+		if (!ASSERT_TRUE(!err || err == -ENOENT, "bpf_map_lookup_batch all cpus"))
+			goto out;
+
+		for (i = 0; i < max_entries; i++) {
+			for (j = 0; j < nr_cpus; j++) {
+				if (!ASSERT_EQ(values[i*nr_cpus + j], j != cpu ? 0 : value,
+					       "value on cpu"))
+					goto out;
+			}
+		}
+	}
+
+	update_opts.cpu = nr_cpus;
+	err = bpf_map_update_elem_opts(map_fd, keys, values, &update_opts);
+	if (!ASSERT_EQ(err, -E2BIG, "bpf_map_update_elem_opts -E2BIG"))
+		goto out;
+
+	err = bpf_map__update_elem_opts(map, keys, key_sz, values, value_sz,
+					&update_opts);
+	if (!ASSERT_EQ(err, -E2BIG, "bpf_map__update_elem_opts -E2BIG"))
+		goto out;
+
+	lookup_opts.cpu = nr_cpus;
+	err = bpf_map_lookup_elem_opts(map_fd, keys, values, &lookup_opts);
+	if (!ASSERT_EQ(err, -E2BIG, "bpf_map_lookup_elem_opts -E2BIG"))
+		goto out;
+
+	err = bpf_map__lookup_elem_opts(map, keys, key_sz, values, value_sz,
+					&lookup_opts);
+	if (!ASSERT_EQ(err, -E2BIG, "bpf_map__lookup_elem_opts -E2BIG"))
+		goto out;
+
+	/* clear value on all cpus */
+	batch_opts.cpu = BPF_ALL_CPUS;
+	batch_opts.elem_flags = BPF_F_CPU;
+	memset(values, 0, max_entries * value_sz);
+	err = bpf_map_update_batch(map_fd, keys, values, &max_entries, &batch_opts);
+	if (!ASSERT_OK(err, "bpf_map_update_batch all cpus"))
+		goto out;
+
+	for (cpu = 0; cpu < nr_cpus; cpu++) {
+		/* update value on current cpu */
+		values[0] = value;
+		update_opts.cpu = cpu;
+		for (i = 0; i < max_entries; i++) {
+			err = bpf_map__update_elem_opts(map, keys + i,
+							key_sz, values,
+							value_sz, &update_opts);
+			if (!ASSERT_OK(err, "bpf_map__update_elem_opts current cpu"))
+				goto out;
+
+			for (j = 0; j < nr_cpus; j++) {
+				/* lookup then check value on CPUs */
+				lookup_opts.cpu = j;
+				err = bpf_map__lookup_elem_opts(map, keys + i,
+								key_sz, values,
+								value_sz,
+								&lookup_opts);
+				if (!ASSERT_OK(err, "bpf_map__lookup_elem_opts current cpu"))
+					goto out;
+				if (!ASSERT_EQ(values[0], j != cpu ? 0 : value,
+					       "bpf_map__lookup_elem_opts value on current cpu"))
+					goto out;
+			}
+		}
+
+		/* clear value on current cpu */
+		values[0] = 0;
+		err = bpf_map__update_elem_opts(map, keys, key_sz, values,
+						value_sz, &update_opts);
+		if (!ASSERT_OK(err, "bpf_map__update_elem_opts current cpu"))
+			goto out;
+	}
+
+out:
+	if (keys)
+		free(keys);
+	if (values)
+		free(values);
+	percpu_array_flag__destroy(skel);
+}
+
 void test_percpu_alloc(void)
 {
 	if (test__start_subtest("array"))
@@ -125,4 +295,6 @@ void test_percpu_alloc(void)
 		test_cgrp_local_storage();
 	if (test__start_subtest("failure_tests"))
 		test_failure();
+	if (test__start_subtest("cpu_flag_tests"))
+		test_cpu_flag();
 }
diff --git a/tools/testing/selftests/bpf/progs/percpu_array_flag.c b/tools/testing/selftests/bpf/progs/percpu_array_flag.c
new file mode 100644
index 000000000000..4d92e121958e
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/percpu_array_flag.c
@@ -0,0 +1,24 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+struct {
+	__uint(type, BPF_MAP_TYPE_PERCPU_ARRAY);
+	__uint(max_entries, 2);
+	__type(key, int);
+	__type(value, u64);
+} percpu SEC(".maps");
+
+SEC("fentry/bpf_fentry_test1")
+int BPF_PROG(test_percpu_array, int x)
+{
+	u64 value = 0xDEADC0DE;
+	int key = 0;
+
+	bpf_map_update_elem(&percpu, &key, &value, BPF_ANY);
+	return 0;
+}
+
+char _license[] SEC("license") = "GPL";
+
-- 
2.50.1